Question of sampler - Githubissues

crowsonkb / k-diffusion

Karras et al. (2022) diffusion models for PyTorch

MIT License

2.26k stars 372 forks source link

Question of sampler #6

Closed UdonDa closed 2 years ago

UdonDa commented 2 years ago

Hi, @crowsonkb

I'd like to ask you the roles of your implemented samples. sample_heun represents a stochastic sampling and sample_lms represents a deterministic sampler, right?

I’m sorry if I misunderstand your implementation and Karras' theory.

crowsonkb commented 2 years ago

sample_heun() is deterministic by default but you can set its s_churn argument to > 0 to do stochastic sampling. sample_lms() is deterministic only. :)

UdonDa commented 2 years ago

Thank you for replying. I missed s_churn! Thank you!!

UdonDa commented 2 years ago

Sorry to bother you again.

You do not implement $S{tmin}$ and $S{tmax}$, right? https://github.com/crowsonkb/k-diffusion/blob/be6254f8e173cd2477e12b448a3375a629ba72a2/k_diffusion/sampling.py#L43

crowsonkb commented 2 years ago

Oh I forgot to, that's all ^^;;

crowsonkb commented 2 years ago

I just committed and pushed this. :)